Vous êtes sur la page 1sur 23

Accounting, Organizations and Society 26 (2001) 327349

www.elsevier.com/locate/aos

Accounting classication and the international


harmonisation debate an empirical investigation
Anne d'Arcy *
Deutsche Borse AG, Listing, Borsenstrae 1, D-60487 Frankfurt am Main, Germany

Abstract
In the current debate on international accounting harmonisation, references are made to accounting model categories. In particular, a distinction is made between the Anglo-American accounting cluster and the continental European cluster. First, we provide a review of the related accounting classication literature. Dierent classication results
can be traced back to dierences in the conceptional and methodological research designs. Second, my own classication attempt is presented which is founded on an actual database of nancial reporting requirements in 14 countries
and the IASC. Basically, a picture of an Anglo-American or continental European accounting model cannot be
established. # 2001 Elsevier Science Ltd. All rights reserved.

In the current debate on international accounting harmonisation, references are made to


accounting model categories. In particular, a distinction is made between the Anglo-American
accounting cluster and the continental European
cluster. The term Anglo-American (or AngloSaxon) accounting is used in the present harmonisation debate with regard to the acceptance of
International Accounting Standards in Europe:
A major problem is identied as being that
the IASC is dominated by the Anglo-American approach to nancial reporting which is
fundamentally dierent from the continental
European approach followed in the EUs
directives. (Flower, 1997a, p. 281)

* Tel.: +49-69-2101-4985; fax: +49-69-2101-1331.


E-mail address: anne_christine_darcy@exchange.de (A. d'Arcy).

Basically, it is possible to identify two groups


who use classication in their argumentation for
or against international accounting harmonisation. The rst group favours the traditional continental European system dominated by the
prudence principle as the Anglo-American
inuence does not take the specic European
environment into consideration (e.g. Hoarau,
1995). Even if they sometimes accept international
rules for listed companies (not all of them do),
they are wary of the impact of international standards on nancial reporting rules for non-listed
companies, especially for tax purposes (Goebel,
1995). The second group prefers Anglo-American
accounting as in their eyes it is better equipped to
inform capital market participants (Epps & Oh,
1997).
Both groups use national accounting clusters as
if their existence has already been proved or generally agreed. Their arguments for or against

0361-3682/01/$ - see front matter # 2001 Elsevier Science Ltd. All rights reserved.
PII: S0361-3682(00)00036-2

328

A. d'Arcy / Accounting, Organizations and Society 26 (2001) 327349

harmonisation could be either supported or questioned by explicit attempts that prove or deny the
existence of the postulated national accounting
clusters. Furthermore, it would be easier to discern
the advantages gained from classifying accounting
systems into models if systems in one cluster had
relatively similar characteristics. To illustrate,
accountants can specialise in national accounting
systems of a specic accounting cluster (Berry,
1987, p. 90), national standard setters may anticipate accounting problems and solutions by looking at the experience of other countries from the
same cluster (Meek & Saudagaran, 1990, p. 154),
or policy-makers may better assess the prospects
and problems of international harmonisation
(Radebough & Gray, 1993, p. 61).
Roberts (1995, p. 641) shows that classications
are neither right nor wrong but useful or less useful for specic research questions. However, the
question remains whether the view of a dichotomised accounting world, implied by certain
authors, is appropriate according to the international harmonisation debate.
When analysing the related literature it is
apparent that explicit classication attempts suggest contradictory groupings. We note that the
dierentiation between Anglo-American and continental European accounting is not statistically
proved. For example, in some classications, the
US and Great Britain are in dierent groups (Da
Costa, Bourgeois & Lawson, 1978; Doupnik,
1987; Frank, 1979; Goodrich, 1982; Nair &
Frank, 1980; Shoenthal, 1989). Furthermore,
Alexander and Archer (2000) seek to establish that
``Anglo-Saxon accounting'' is a myth by critically
examining four putative commonalties which are
frequently attributed to the UK and USA
approaches to nancial accounting.
As a result, the use of these classications in the
international harmonisation debate may be questioned. This paper analyses whether there is a
contradiction between the intended use and the
scientic ndings and shows possible consequences for the political process of accounting
harmonisation. Basically, the question of whether,
and under which conditions, an Anglo-American
and a continental European accounting group can
be identied needs to be resolved. Moreover, this

contribution seeks to improve the methodology


and update the data compared to other prominent
cluster studies.
One reason for the dierences in the accounting
clusters could be due to the diverse characteristics
of accounting systems (Roberts, 1995, p. 660).
Many criteria seem to be relevant for determining
or rather describing them. Thus, depending on the
attributes selected, dierent clusters of nations in
accounting systems may result. Obviously there
are dierent levels of reasoning to which the
authors refer. In the intended analysis these levels
of reasoning are described. They lead to the specic research design concept that can be distinguished in the study matter and the choice of
the classication criterion. One level of reasoning
could be the actual reporting requirements, but
this has not yet been used for classication purposes (Rahman, Perera & Ganeshanadam, 1996,
p. 338). Thus, a classication attempt should be
founded on an actual database of nancial
reporting requirements in 14 countries and the
IASC (Ordelheide & Semler, 1995). This attempt
could oer new insights regarding the harmonisation debate.
There is a fair amount of choice regarding
the methodological proceedings. The aim of
this analysis is to explain dierences in the proposed classications. Classication methods are
heuristic and can be characterised as ``rules of
thumb'' (Aldenderfer & Blasheld, 1984, p. 14).
Consequently, in the following analysis the main
focus is the validation of the cluster solutions
in varying hierarchical cluster methods and the
comparison of the results with a solution using
multidimensional scaling.
1. Conceptional research design: study matter and
classication criterion
The search for a theoretical basis for a classication of national accounting systems is closely
related to the specic study matter and additionally to the classication criterion. Although the
choice of a specic classication criterion is at
rst glance a technical aspect of the research
design, it implicitly or explicitly denes the

A. d'Arcy / Accounting, Organizations and Society 26 (2001) 327349

specic attributes of the phenomenon ``accounting


system''.
In the following, some of the various classication attempts will be briey reviewed while paying
particular attention to the classication criterion
and the study matter. Regarding the various
approaches of classication, it is immediately
clear that the authors argue on dierent levels of
analysis (Bloom, Long & Collins, 1994, p. 265).
These levels become obvious when one looks at
the dierent attributes chosen for classication
purposes.
1.1. Cultural areas and accounting
The rst level of analysis is the cultural reasoning. At this level the hypothesis is that culture has
a signicant inuence on accounting (cultural
determinism theory or cultural cause theory;
Belkaoui, 1994, p. 51; Riahi-Belkaoui, 1997, p. 17)
implying that cultural groupings correlate with
classications of national accounting systems.
Overall, any cultural-related study links cultural
areas to accounting system characteristics (e.g.
Fechner & Kilgore, 1994; Gray, 1988; Hofstede,
1980; Perera, 1989; Perera & Mathews, 1990) or to
the structure of standard setting processes (Belkaoui, 1989; Bloom & Naciri, 1989; Boczo, 1996;
MacArthur, 1996). One problem in this argumentation is the subjective choice of elements
which are used to dene cultural areas. We do not
know whether Hofstede's four dimensions or
other factors used by other authors are sucient
and exhaustive to describe culture (Hofstede &
Schreuder, 1987, p. 30). In addition, all empirical
research in this eld is based on data which focuses
only on a small part of the ``accounting world''
(Montagna, 1987, p. 2325) because the input of
questionnaires are used.
The second problem is the assumption that
accounting can be described as a function of culture (Bloom et al., 1994, p. 265). It is apparent
that there is a relationship between cultural peculiarities and national accounting systems. But it is
questionable whether societal values alone can
explain accounting developments. Basically, we do
not know which aspects of accounting are strongly
aected by cultural inuences, which aspects are

329

less aected and which aspects are hardly eected


at all (Baydon & Willett, 1995, p. 83). The authors
only oer plausible examples of the connection
(Gray, 1988). However, examples that imply
counter evidence can easily be found (Baydon &
Willett, p. 83).
To summarise, it may be possible that there is a
link between cultural groupings and the resulting
accounting models but it is not possible to verify
the culture-based hypothesis. Therefore, other
lines of reasoning have to be found to prove the
existence or the usefulness of positing an AngloAmerican accounting or continental European
accounting cluster for the harmonisation debate.
1.2. Environmental groupings and accounting
According to the Environmental Determinism
Theory, it is assumed that there is a close connection between environmental factors and accounting systems. The most frequently mentioned
environmental factors found in recent publications
are the legal system, the providers of capital, the
tax system, the inuence of the accounting profession, and the importance of nance and capital
markets (Alexander & Nobes, 1994, pp. 6775;
Radebaugh & Gray, 1993, p. 44). Also there are
inuences like standards (Peller & Schwitter, 1991,
p. 4.5) or the political system (Choi & Mueller,
1992, pp. 4043).
The rst authors to write about the likelihood of
environmental groupings (AAA 1977; Mueller,
1967, 1968; Previts, 1975; Seidler, 1967), used
these factors intuitively to create clusters. Typically, sets of factors are proposed which serve as
an explanation of internationally observed dierences in prevailing accounting principles or practices. The link between these factors and national
accounting systems is merely described and not
analysed further or tested empirically. More
recently, major international accounting textbooks
provide a list of environmental factors and
describe possible connections to the peculiarities
of national accounting systems (Choi & Mueller,
1992; Nobes & Parker, 1995; Haskins, Ferris &
Selling, 1996).
Based on the outlined descriptive analysis,
empirical researchers in the area of international

330

A. d'Arcy / Accounting, Organizations and Society 26 (2001) 327349

accounting aim to validate the assumed dependencies (Cooke & Wallace, 1990; Nair, 1982; Nair
& Frank, 1980). To reduce complexity, some of
the approaches investigate the relationship with
respect to a specic accounting problem (Needles,
Powers & Revsine, 1991) or a single factor
(Doupnik & Salter, 1995; Shoenthal, 1989), or
focus on a narrow selection of national environments (Cooke, 1992; Hagigi & Sponza, 1990).
The most sophisticated and most frequently
cited approach in this research area is that of
Nobes (1983, 1992). His empirical results prove
the predicted relationship. But the core assumptions on the direction and intensity of the inuence, i.e. the subjectively determined scores, are
derived from the hypothesis to be tested (Doupnik
& Salter, 1993, p. 44). Despite this potential
objection, the research design has been copied and
referred to ever since (AlNajjar, 1986; Doupnik &
Salter, 1993, 1995).
Although most authors agree with the environmental view, there are several, partly contradictory groupings. Coming back to the original
question, whether an Anglo-American or continental European accounting cluster can be identied, the existing cluster attempts do not provide
us with a straightforward answer. In contrast to
the expectation, the ndings of most investigations
suggest separate UK and US groups (AAA, 1977;
Frank, 1979; Mueller, 1968; Nair & Frank, 1980;
Seidler, 1967; Shoenthal, 1989). A more dierentiated view provides Nobes (1992) and Doupnik
and Salter (1993, 1995) with a hierarchical solution. It is only in the two-cluster-solution that the
UK and US are both grouped in one cluster.
The problem arises because the choice of relevant elements, which should separate environmental areas, is as subjective as the methods of
measurement and the explicit relation to accounting systems. The list of important environmental
factors proposed by the descriptive approaches
are seldom used by the empirical researchers.
Mathews and Perera (1991, p. 305) conduct an
analysis of factor selection in six prominent papers
and show that they are far from any general
agreement about the main environmental factors.
Additionally, one has to take into account that
there are not only national environments but

also within each country dierent enterprises may


have dierent environments (van der Tas, 1995,
p. 257).
Moreover, the direction and intensity of the
factors' inuence on an accounting system remains
unclear. In a black box view, the mechanism of
causation is not addressed. Consequently, certain
characteristics of accounting systems cannot be
explained by environmental conditions (Bockem
& d'Arcy, 1999, p. 66).
Yet, even if we accept a causal dependence,
another problem arises. If we interpret the
accounting system as causally aected by the
national environment, the specic national system
evolves as the ecient solution. Evolutionary economics is characterised in turn by the fundamental
assumption of the survival of the ttest (SOTF);
with regards to accounting system classication
this implies that the ttest (most ecient)
accounting system survives in a nation specic
factor environment (Watts & Zimmerman, 1986,
p. 8). Critics of the SOTF hypothesis mainly focus
on the fact that the underlying institutional reasoning is not suciently specied and lacks explanatory power in important instances (Terberger,
1994, p. 140). The ndings of game theory
demonstrate that inecient equilibria are also
possible (Bockem & d'Arcy, 1999). Thus, it is not
possible to deduce the accounting system from
environmental factors in a scientic sense. Otherwise, this form of ``lawlike generalisation'' would
ignore the dierence between natural phenomena
and accounting as a human activity (Ryan, 1985,
p. 87).
As for the cultural level we can draw the same
conclusion that sets of environmental conditions
could account for the existing accounting models.
However, they do not have enough explanatory
power to support the hypothesis. It does not seem
appropriate to use these explanations alone for the
harmonisation debate.
1.3. Types of the standard setting process
The national institutions and processes of standard setting can be dened as either a cause of
dierences or as an integrated part of national
accounting system. On the one hand, the process

A. d'Arcy / Accounting, Organizations and Society 26 (2001) 327349

and the bodies inuence accounting standards


and practices. Thus, one reason for the dierences
can be found in these various standard setting
activities (Daley & Mueller, 1989, p. 22). On the
other hand, the ``approaches used to establish
accounting standards are a function of cultural
values and factors inherent in each country.''
(Bloom & Naciri, 1989, p. 70). Therefore, several
investigations analyse and compare the national
standard setting institutions and processes in various countries as part of the accounting system
and explain dierences by environmental factors
(Puxty et al., 1987; Taylor & Turley, 1986).
Besides an explicit classication by ``rule making''
provided by Nobes (1992, pp. 99103), some
authors identify several types or baselines of standard setting (Bloom & Naciri, 1989; Day, 1996).
With reference to our original question, it is
astonishing that an Anglo-American cluster of
standard setting cannot be identied whereas a
continental European one can. The US system
seems to be placed in between the clusters of
continental Europe and some Commonwealth
countries.
These groupings of standard setting systems are
not so dierent from environmental clusters. So, it
is obvious that there is a relationship between
these factors and the various types of standard
setting processes. But whether the accounting
practices are directly inuenced by these factors or
by the process cannot be determined unambiguously. Additionally, one must take into account
that these institutional dierences are not necessarily congruent with the standard setting process
at an informal level (Ordelheide, 1997).
Altogether, an isolated analysis of the standard
setting process is not sucient to dene or explain
dierent accounting models. However, for the
harmonisation debate this disharmony of standard
setting approaches should be taken into account,
because the process can be understood as one part
of a national accounting system (Flower, 1997b,
p. 48).
1.4. Classications of accounting practices
There are many investigations which analyse
dierences in accounting practices inter-

331

nationally.1 All these attempts show serious differences in accounting practices. The rst
approaches using statistical methods to cluster
national accounting systems according to reporting practices are found in the late seventies (Barrett, 1977).2 The publication of three databases on
accounting practices by Price Waterhouse (1973,
1975, 1979) was the driving motive for several
researchers to use factor analysis to nd national
accounting clusters. In 1978 Da Costa et al. tested
the posited existence of groupings of accounting
models using this database. They were followed by
Doupnik (1987), Frank (1979), Goodrich (1982),
Nair (1982), Nair and Frank (1980) and Nobes
and Matatko (1980). In the nineties, other
empirical studies with partly new databases test
the relationship between accounting practices
classication and certain environmental inuences
(Nobes 1992; Salter & Doupnik, 1992; Doupnik &
Salter, 1993, 1995).
The results are as non-homogeneous as at other
levels already mentioned. However, for all the
investigations which use the Price Waterhouse
surveys as a database, two dierent clusters for
UK and US can be distinguished.
In addition to the problem that there is no consensus as regards national accounting clusters,
several limitations characterise the attempts on
this level. In the last section, we saw that dierences do not occur only in an international but
also in an intranational context (Archer et al.,
1995, p. 67). Furthermore, temporal events can
inuence the cluster result. Unsystematic inuences such as cyclical uctuations or management
incentives may adulterate the ndings. Nobes
1
First, some approaches examine the actual impact of differences in income or equity by comparing the national numbers from various enterprises with the amounts after a
reconciliation to US-GAAP (Amir, Harris & Venuti, 1993;
Weetman & Gray, 1991). Others try to measure the dierences
by indexes as an indicator for harmonisation eorts (Archer,
Delvaille & McLeary, 1995; Emenyonu & Gray, 1992; Herrmann & Thomas, 1995) or other approaches (Joos & Lang,
1994; Walton, 1992).
2
It is possible to see a plain bipartition in intuitive and
statistical methods of analysis: until 1977, ending with the
AAA, only intuitive methods can be found, since 1977, starting
with Barrett, statistical techniques are applied.

332

A. d'Arcy / Accounting, Organizations and Society 26 (2001) 327349

(1983, p. 5) is critical of the hypotheses. The clustering of the Price Waterhouse data should be
used to ``generate theories'', not to prove them
(Nobes & Matatko, 1980, p. 11). Several practical
problems also exist, e.g. answers are needed for
the following questions: what do accounting practices entail and how can these practices be carried
out? All the databases are founded on the analysis
of several annual reports. Consequently, the
results are biased towards the specic enterprises
chosen for data collection (Nobes, 1983, p. 3).
Further critiques of attempts using accounting
practices for cluster purposes concentrate on the
Price Waterhouse surveys. In order to create an
adequate database on this level for classication
purposes, the results of the attempts should be
interpreted with caution. However, one advantage
of classications using accounting practices is that
they provide us with new insights into international accounting dierences and thus consequently serve as a useful source of information for
the harmonisation debate.
1.5. Clustering nancial reporting requirements
It is astonishing that no one has attempted to
classify national accounting systems by nancial
reporting requirements. Of course, there are several investigations which collect or analyse the
dierences in accounting standards in several
countries (Gray et al., 1984; Ze, 1972). Yet none
of them build clusters based on their ndings. One
reason could be that a database, such as the one
provided for accounting practices by the three
surveys of Price Waterhouse, does not exist. Rahman et al. (1996) analyse the congruency of nancial reporting requirements in order to learn
something about ``formal harmonisation'', but
only for the accounting systems of Australia and
New Zealand (Rahman et al., p. 338).
The term ``nancial reporting rules'' has to be
dened rst. As we know from the classication
approaches based on standard setting processes,
dierent processes generally imply dierent types
of accounting rules. Nevertheless, the functions of
these dierent types of rules are similar, namely
the control of the accountant's behaviour. A system of accounting rules explicitly or implicitly

forbids, allows or requires certain accounting


methods independently of its legal basis. Consequently, nancial reporting rules are those statements that require, forbid or allow certain
accounting methods. In the following, only these
rules which apply to corporations will be taken
into account. Additionally, only the rules which
eect quoted companies are considered. This limitation is necessary as the harmonisation debate
relates to listed companies, only.
A de jure classication by nancial reporting
requirements oers some advantages. On one
hand this level seems exible enough to reect
important developments like the EU-Directives.
On the other hand, it shows sucient continuity
to avoid the impact of unsystematic temporal
inuences. Compared with a de facto classication
of accounting practices, it rejects possible intranational eects. A more powerful argumentation is
that the de jure harmonisation takes place on this
level. It is only possible in the rst place to harmonise standards, not environmental or cultural
factors nor accounting practices for all listed
companies in one country. Accordingly, tests must
be carried out to show whether an Anglo-American cluster or indeed others like the continental
European cluster can be identied on this level.
Only then can it be decided, if harmonisation is
either impossible or possible only with regard to
the environmental conditions.
In addition, one inconvenience of the environmental-based classications can be eradicated.
One of the core criticisms of the environmentalbased classication attempts was the insucient
specication of the dependency between the
accounting system and the environmental factors.
To throw light into the dependency black box, an
individual-based preference analysis is preferred
instead of making factors the central unit of analysis. This perspective is in line with the widely
accepted methodological individualism which was
labelled as the ``actors-for-factors-view'' by
Bockem and d'Arcy (1999, p. 67). This ``actors-forfactors-perspective'' leads to an interpretation of
national accounting rules as a compromise solution which is induced by a mechanism for aggregating the actors' preferences and the national
preferences (namely the national standard setting

A. d'Arcy / Accounting, Organizations and Society 26 (2001) 327349

process). The actors' preferences are in turn aected by actors' incentive structures which are inuenced by various environmental factors (Bockem
& d'Arcy, p. 71). Consequently, national accounting rules reect environmental factors in the
specic outcome of the actors' incentives and preferences and allow the accounting systems to be
interpreted as human activity systems.
As with the other approaches, we are aware of
the limitations. Firstly, the choice of the variables
used to separate clusters is still subjective. Dierent choices would result in dierent cluster solutions. Secondly, the intensity and eectiveness of
nancial reporting in a country is a function not
only of the reporting requirements but also of the
degree of enforcement. Consequently, it is possible
that an accounting method is required but not
used, or that it is forbidden but possible to nd in
some nancial statements. Or a method may be
allowed but it may be used often or hardly at all.
Moreover, accounting practices require judgement
and there may be dierences in practices between
national accounting systems. Additionally, there
may be dierent rules for enterprises within a
country and/or the same rules for enterprises in
various jurisdictions. However, we are able to
resolve the latter problem if we concentrate purely
on the rules for listed companies. But we must be
aware that we only compare national accounting
rules for listed companies and not rules regarding
all types of companies. This can be justied by the
core arguments of the harmonisation debate that
concentrate on rules for listed companies. However, even if the use of international rules for listed
companies are accepted another debate may
explicitly state the negative impact of this on
nancial reporting rules for non-listed companies,
especially for tax purposes. Finally, it is also possible to argue that de facto harmonisation is more
relevant than the de jure harmonisation that is
followed in this paper. The harmonisation eects
of IASs and international capital market consideration typically make themselves apparent
through changes in practices before changes in
national requirements. But still a de jure harmonisation is necessary to force a de facto harmonisation for all companies not only on a voluntary
basis.

333

As shown in the previous sections, it is apparent


that every classication criterion will lead to practical and methodological problems and limitations. One has to consider that every construction
of groupings reects the author's own prejudice
about the discipline (Roberts, 1995, p. 641). However, the nancial reporting regulation basis provides the possibility to gain new insights into
clusters of national accounting systems because it
is established on the same level at which the harmonisation takes place. As a consequence, the new
classication attempt provided here can be understood as a complementary approach which may
provide additional information on some of the
characteristics of national accounting systems.
2. Methodical research design
The most statistical approaches are based on the
surveys from Price Waterhouse (1973, 1975, 1979).
These surveys provide an overview of disclosure,
valuation, and recognition practices in 37, 46, or
rather 64 countries using a 7-step (for 1973 6-step)
ordinal scale. They were based on what Price
Waterhouse oces reported as being the principles
and practices of the reviewed national accounting
systems. General accounting principles, such as
the realisation concept, as well as special cases,
such as the exclusion from consolidation of a
subsidiary with dissimilar business, are presented.3
There are some mistakes in the data (Nobes,
1981, p. 268) and it is possible that the reports
from non-representative enterprises are included
(Nair & Frank, 1980, pp. 444445). The data collected is not suitable for statistical classication
purposes because the information is qualitative,
sometimes ambiguous, and not weighted so that
important parts are swamped by trivial ones
(Nobes & Matatko, 1980, p. 70). Additionally, the
problem of exaggerating the dierences between
the US and UK arises because the compilers of the
survey questions are most familiar with the
accounting system in these countries (Nobes, 1983,
3
The seven categories for the 1979 study are: Required,
Insisted upon, Predominant practice, Minority practice, Rarely
or not found, Not accepted, Not permitted.

334

A. d'Arcy / Accounting, Organizations and Society 26 (2001) 327349

p. 2). Furthermore, the database mixes up information on accounting rules and practices, because
categories that include both types of information
such as ``rarely or not found'' and ``not permitted''
are used. This results in dierent aggregation
levels. Finally, the information is outdated, for
example the changes relating to the 4th, and 7th
EU-Directives are not included.
Some of the aforementioned problems also arise
with the other used databases. In particular, the
choice of variables used by AlNajjar (1986), Hofstede (1980) or Nobes (1992) are also subjective
and aected by their socio-economic environment.
Consequently, a self-generated database does not
immunise against these inuences. Furthermore,
subjective scoring systems are not suitable to test a
hypothesis because the same assumptions as for
the hypothesis are used. Additionally, Nobes and
AlNajjar use criteria on dierent aggregation
levels, namely causal factors and measurement
practices.
Except for the work of Doupnik and Salter, all
other statistical classication approaches use
databases that are ordinally scaled. Hence, the
data must be transformed because most statistical
methods allow the input of nominal or metric
(cardinal) data, only. Accordingly, some authors
convert the ordinal data into metric or use the
ordinal categories as metric numbers directly. This
renders statistical analysis such as the calculation
of averages possible, although the data foundation
is not suitable for this purpose.
It is notable that the criticism of certain classication attempts concentrates on the scope of analysis and the databases but never on the statistical
procedures.4 This is astonishing because consensus
in the statistical methods to be used cannot be
found at all. A relatively sophisticated and wellknown procedure to nd accounting clusters is
factor analysis. Basically the clusters are built
based on the highest rotated factor scores (Eigenvalue>1) so that there are as many clusters as
factors. Each item (country) is attributed to a
4
Even though Nobes (1983, p. 5) criticises the application of
factor analysis because this method does not produce hierarchies, he does not discuss the main problems which result
when clusters are found.

cluster which stands for the factor for which this


country has the highest factor loading. As a result,
the groups are not built based on the similarity of
various variables or factors, but on the similarity
according to one factor only. Consequently, the
national nancial reporting systems are symbolised by only one, supposedly ``typical'' factor
irrespective of whether other factors are important
or not. Additionally, it is not possible to conclude
how close some groups are. Thus, the applied
methods constitute a serious limitation for these
approaches and for their results (Archer &
McLeay, 1992, p. 1).
On the contrary, AlNajjar (1986), Doupnik and
Salter (1993, 1995), Nobes (1992) and Salter and
Doupnik (1992) utilise hierarchical cluster
approaches, like the combination from AverageLinkage, Furthest Linkage, and Ward to nd
national accounting groups. The hierarchical
structure of the classication shows associations,
which allow further interpretation. Accordingly,
cluster analysis seems to be appropriate to nding
accounting clusters.
2.1. Actual study
The sample for my empirical study comprises 15
national systems, including various European
countries, and the US, Canada, Australia, and
Japan. The sample also includes the rules of the
IASC. I am aware that the rules of the IASC are
not well suited to the idea of comparing national
accounting systems. However, the IAS are often
perceived as conforming to the Anglo-American
accounting model. To test this presumption it is
necessary to include the rules of the IASC into the
data base. All results presented below are still
valid regardless of whether the IAS are included or
excluded from the database.
The data basis applied is based on the TRANSACC Reference Matrix (Ordelheide & Semler,
1995). This matrix is part of the Transnational
Accounting project (TRANSACC) (Ordelheide &
KPMG, 1995) that systematically sets out the
nancial accounting rules and practices of several
countries. Accounting professors and professionals describe accounting problems and the
related solutions for their respective countries in a

A. d'Arcy / Accounting, Organizations and Society 26 (2001) 327349

335

Table 1
Content of the current database
P

Topics

Example

Individual accounts
Recognition assets
Recognition liabilities
Valuation assets
Valuation liabilities
Revaluation accounting

Recognition of development costs


Provision for expenditure
LIFO
Valuation of foreign currency liabilities
Treatment of revaluation reserve

Group accounts
Full consolidation set
Proportional consolidation set
Uniformity of accounts
Foreign currency translation
Consolidation of capital
Consolidation of debt/prots
Equity method
Deferred taxation

Exclusion in cases of diverging activities


Rights which assure joint control
Uniform accounting policies
Treatment of translation adjustment
Pooling of interest method
Treatment of dierences
Exercise of signicant inuence
Netting of tax assets and liabilities

systematic way. The rst of the two chapters on


each country explains the accounting problems
that aect only individual accounts. The second
deals with the additional problems arising from
the preparation of group accounts. The information on individual countries and the IASC is summarised in the Reference Matrix, so that each
country's complete accounting rules are presented
in tabular form and the rules on any particular
accounting problem can be read for all the countries covered by TRANSACC. Accordingly, the
matrix covers those rules that determine the content of the balance sheet and of the prot and loss
account, including recognition and valuation as
well as consolidation methods carried out in the
respective countries. It provides information on each
accounting method under review of the following
form:5 R (required) A (allowed) F (forbidden)6
The content of the current data base can be summarised as in Table 1.
As a next step, I dichotomise the data. It is
necessary to change from an ordinal to a nominal
scaling degree without losing information and at
the same time avoiding interval assumptions. An
example may be better equipped to illustrate this
procedure: the item ``recognition of development
costs'' can be dichotomised in the two questions
``recognition required'' and ``recognition forbidden''.

Variables after coding


P
71
20
7
37
4
3
P
58
16
3
6
7
13
2
8
3

Correspondingly, with the coding 1, 0 for


required, 0, 0 for allowed and 0, 1 for forbidden all
possible entries can be expressed.
The hierarchical classication methods seem to
be appropriate for my research purpose. A cluster
analysis is carried out in several steps. The initial
step after coding is the calculation of similarities
or dierences between each pair of the analysed
national systems. In the following, the Simple
Matching Similarity Coecient (SM) is used,
because the results can be easily interpreted
(Sneath & Sokal, 1973, p. 146). The SM measures
the relative (percentage) congruence between two
national systems. Thus, matches and mismatches
of all analysed accounting methods are equally
weighted. The closer the SM reaches 1 or 100%

5
Details about how this data are derived are available on
request from the author and are reported in d'Arcy (1999).
6
The symbols represents the following (Ordelheide & Semler, 1995, p. 4):

R: the accounting method indicated is required by law or


is highly recommended by the standard setter
A: the accounting method may be used but is not compulsory or is not recommended by the standard setter
F: the accounting method is forbidden or corporations
are discouraged to apply it.

336

A. d'Arcy / Accounting, Organizations and Society 26 (2001) 327349

the more comparable are the two national


accounting systems.
In the following, dierent clustering techniques
for the classication of binary data sets are
applied in order to nd cluster solutions of high
stability (Gordon, 1993/1981, p. 121). Generally,
all hierarchical agglomerative methods have this in
common: the rst pair is built in regard to the top
similarity coecient (or the minimal distance
coecient). Then the remaining coecients are
recalculated using the data obtained from the new
cluster. The algorithm for the recalculation varies
as follows (Gordon):
1. average linkage between groups: the average
of each group determines the similarity coefcient relevant for fusion;
2. single linkage (nearest neighbour): the nearest neighbour of each cluster determines the
similarity coecient relevant for fusion;
3. complete linkage (furthest neighbour): the
furthest neighbour of each cluster determines
the similarity coecient relevant for fusion;
and
4. average linkage within groups: the average of
the cluster to be built is maximised.
The conceptional and methodical research
design of the main classication attempts and this
study can be summarised as in Table 2.
3. Results
3.1. Simple matching similarity coecient (SM)
The top SM value and thus the highest congruence show Germany and Austria with 91.47%.
This outcome is not surprising, because the Austrian accounting system is based on the German
one (Lukas, 1995, pp. 10211022). The highest
distance can be found between France and the US
with 29.46% congruence only. This nding is
almost comparable with the outcome of other
cluster studies. Although these two coecients
show great variety, most values are situated in a
medium interval: more than 70% of the combinations can be found in the interval that shows more

than 50% and up to 70% congruence. Consequently, most pairs are neither extremely homogeneous nor heterogeneous.
Concentrating on high distances besides France
and the US, the results derived from the Australian system are remarkable. Only the IASC
shows a coecient close to the average. All other
pairs including Australia are measured with less
than 60% congruence. Correspondingly, the Australian system seems to have an outsider position.
Moreover, the ndings imply that statements,
which refer to the comparability of two systems,
should only be interpreted with caution. For
example, an isolated analysis of Canada and the
US shows that both systems with 67.44% congruence reach the highest value for each other,
meaning that for Canada the US system is the
most comparable and vice versa. But taking all
national systems into account, there are 15 pairs
that lie in an interval of higher congruence.
3.2. Cluster solutions
As the rst method the Average Linkage
between Groups is chosen because it combines
certain characteristics of the Single and the Complete Linkage. Fig. 1 shows the dendrogram for
the Average Linkage solution.
This dendrogram illustrates the fusion process
and should be read from the top left to the bottom
right. The numbers are the standardised similarity
coecients (fusion distances) that are relevant for
each fusion.7 The rst cluster is Germany and
Austria as expected. Belgium and France follow it.
These four national systems form a clear cluster,
which could be characterised as a continental
European group. This group is enhanced by other
European systems such as Switzerland, Denmark,
the Netherlands and the UK, so that a European
group emerges. A second, more heterogeneous
group with Spain, Japan, and Sweden merges with
the rst group. Canada, US and the IASC form a
third group, approximately at the fusion level of
7
The applied Software SPSS standardizes the fusion distances from 1 to 25. The rst fusion, in the present contribution
Germany and Austria, gets the fusion distance 1, the last fusion
the fusion distance 25 (Norus is, 1994, p. 91).

Table 2
Survey of some characteristics of the main classication attempts
Study

Conceptional research design

Methodical research design

Findings
P

Countries

Classication criterion

Database

Level of aggregation Scaling level

Method of analysis

Cluster

Hateld (1911)
USA

Describing accounting dierences


between UK/FRA/GER/USA

Financial reporting
practices

Self-generated

Intuitive

USA, UK, continental


European (GER, FRA)

Seidler (1967)
USA

Develop and present coherent


patterns of international
accounting variation

Spheres on inuence
based on economic,
political and legal
factor

Self-generated

Intuitive

None

British, US,
continental-European

Mueller (1967)

Find fundamental patterns of

Importance of economic,

Self-generated

Intuitive

None,

Patterns of development:

USA

accounting development as a
framework for analysis

governmental, business
factors in the development
of accounting systems

Mueller (1968)
USA

Classication based on national


business environments

States of economic
development, of business
complexity, shades of
political persuasion,

examples
for each
group

Self-generated

Intuitive

None,
exampIes
for each
group

macro-economic (SWE),
microeconomic (NETI-I),
independent discipline
(UK, US), uniform (GER)
10

Sets of business
environments, e.g.
US/CAN/NETH, British
Commonwealth (without

reliance on some
particular system of law

CAN), GER/JAP

Previts (1975)
USA

Classication according to positive, Environmental


normative and historical factors
conditions, future
inuence of standard
setters, historical
circumstances

Self-generated

Intuitive

AAA (1977)

Find international zones of

Eight parameters for

Self-generated

Global

Intuitive

USA

inuence, propose a morphology

inuencing factors

Barrett (1977)
USA

Empirically test whether the


extent of nancial disclosure
in foreign annual reports is
signicantly dierent from that
found in US reports

Disclosure practices
(17 items of
information drawn
from annual reports of
103 rms 19631972)

Self-generated

Uniform

Ordinal
! cardinal

Da Costa et al.
(1978) Canada

Test the posited existence of


groupings of accounting models,

Accounting practices
(100 out of 233)

PW 73

Various

Ordinal ! ? Principal
(not specied, component

search for environmental factors

Results, cluster denition

None

2
(as
hypothesis)

Anglo-American,
Continental

6 as

British, FRA/ES/POR,

example

probable
cardinal)

Weighted index

GER/NETH, communistic

US/UK, JAP/SWE/NETH/
GER, FRA
(others not classied)

38

UK (former British empire)


versus US model

A. d'Arcy / Accounting, Organizations and Society 26 (2001) 327349

Study topic

factor analysis
+ Q-analysis
based on sum
scores on each
factor, for each
country

337

(Table continued on next page)

338

Table 2 (continued)
Study

Frank (1979)
USA

Conceptional research design

Methodical research design

Findings

Study topic

Classication criterion

Database

Level of aggregation Scaling level

Method of analysis

Examine the extent to which


dierent patterns of accounting
concepts and practices exist and
relate dierences to economic

Accounting practices
(233)

PW 73 +
self-generated

Various

Principal component
factor analysis,
grouping based on
highest rotated

Ordinal
! cardinal
(% of usage)

and environmental factors

Countries

38

Examine dierences in
classications whether
measurement or disclosure
practices are used + the
association with underlying

Accounting practices
dierentiated in
measurement
(mp=147/162)
+disclosure

environmental variables

practices
(dp=86/102)/
environmental
factors

PW 73+75+
self-generated

Various

Ordinal
! cardinal
(% of usage)

Principal
component
factor analysis,
grouping based
on highest rotated

PW 73+75

Various

Nair (1982)

Develop empirical guidelines

Accounting

PW 79

Various

USA

for comparing nancial


accounting data + replicate
earlier study by Nair/Frank
for examine the stability of
country groupings over time

practices
dierentiated in
measurement
practices (mp) and
disclosure
practices (dp) (267)

38/46

Goodrich (1982)

Attempt to delineate factor

Accounting

UK

groupings based on PW-data

principles
and policies (26)

Doupnik (1987)
USA

Grouping countries according


to their respective degrees of
commonality on two points of
time in order to measure the
harmonisation eorts since the

Accounting
practices
(70)

PW 79

PW 75 +
self-generated

Uniform

Various

Factor analysis
(following DBL) +
Q-Analysis

38

64

Ordinal

Principal component
factor analysis,
grouping based on
highest rotated
factor scores

Ordinal

Factor analysis,

(Kendall
correlation
coecient)

grouping based on
highest (oblique)
rotated factor scores

Ordinal
! cardinal
(scores)

Factor analysis
(oblique rotation)
(after Nair/Frank)

British-Commonwealth,
Latin American, Continental
European, US-model,
indirect support of

4/5 (mp),
7/7 (dp)

mp 1975: BritishCommonwealth, Latin


Americanl South European,
Northern and Central
European, US model, Chile;
support of environmental
determinism theory for mp
only

Ordinal
! cardinal

! cardinal
(% of usage)

Results, cluster denition

environmental determinism
theory

factor scores +
multiple discriminant
analyses

Accounting
practices
(233)

64

12

Support British model only

7 (mp),

Mp: British-Commonwealth,

10 (dp)

Central America + Southern


Europe, Central Europe +
Scandinavia, US, South
American, African, India

Prototypes: US, Switzerland,


UK, Brazil, Jersey

36

establishment of the IASC


in 1973
Propose a general model for
dierent modes of accounting
regulation

Cluster

factor scores

Nobes and Matatko Test the validity of the DBL


(1980) UK
1978 classication with
varying number of factors

Puxty et al.
(1987)

For 83-Analysis: British


Commonwealth, South
American/South European,
JAP/Panama/US, CAN/
MEX, Scandinavia, GER;
dierences have decreased
and quality has improved

Accounting
standard
setting

Self-generated

Intuitive

None,
examples
for each
group

Four modes of regulation:


4 (and
combinations) Liberalism,
Associationism
(UK), Corporatism
(SWE), Legalism (GER)

A. d'Arcy / Accounting, Organizations and Society 26 (2001) 327349

Nair and Frank


(1980) USA

Daley and Mueller


(1989) USA

Typify the standard setting


process according to the role
of private and governmental

Accounting
standard
setting

Self-generated

Intuitive

institutions in the
development and
enforcement of accounting

Nobes (1992)
UK

AlNajjar (1986)
USA

Purely political approach


(GER, FRA), private
professional approach

group

Test the validity


of proposed hierarchical
classication with a self-

Two explanatory
variables +
seven measurement

generated morphology
and PW-data

practices,
accounting
practices (267)

Find classication of
standard setting processes

Accounting
standard

and institutions

setting

Test the validity of a


classication proposed
by Nobes + own hypotheses
about an international

Five causal
factors, 11
standardisation
elements

Self-generated

Various

Ordinal !
cardinal
(scores 15)

Cultural value
dimensions

Hofstede 80

Uniform

Cardinal
(scores)

Self-generated + Various
PW 79

Ordinal
! cardinal
(scores 14)

Cluster analysis by
nearest neighbours
and furthest

(Australia, CAN, UK),


public/private mixed
approach (US), broadly
mixed approach (NETH)

14, 64

Various
because
hierarchical

Support hypotheses
(but few deviations)

Various
because

2-cluster solution:
non-government creation

hierarchical

(US, Australia, CAN, UK,


NETH, New Zealand),
government creation
(GER, JAP, FRA)

Clustering by nearest + 8
farthest neighbour on
total dierences
between countries

Various
because
hierarchical

Support hypotheses
(but few deviations)

Cluster analysis and


multidimensional
scaling (adopted
from Hofstede

50

Adopt Hofstede's culture


area classication

Support hypothesis for


UK and US

neighbours based
on roots of the
squared dierences,
dierent methods for
PW-data
Self-generated

Intuitive

tendency toward more


standardisation and increasing
governmental jurisdiction
Gray (1988)
UK

Explore the extent to which


international dierences may
be explained by dierences
in cultural factors, link
accounting values and
culture area classications

Shoenthal (1989)
USA

Determine whether the


education factor is an
essential element in
clustering, test for UK
and US only

1980)

Accounting
competencies
(63)

Self-generated

Uniform

Ordinal
Discriminant analysis
(dichotomised)
+ cardinal

339

(Table continued on next page)

A. d'Arcy / Accounting, Organizations and Society 26 (2001) 327349

Nobes
(19831994)
UK

None,
examples
for each

340

Table 2 (continued)
Study

Conceptional research design


Study topic

Cooke and Wallace Test whether the level


(1990) UK
of corporate nancial

Methodical research design

Findings

Database

Disclosure
regulation

PW 79,
Uniform
Gray/Campbell/

Ordinal !
cardinal

Index scores and


Moses test of extreme

Shaw 84

(scores)

reaction

Modied PW 79 Various
+ self-generated

Cardinal

Cluster analysis
(centroid + ward)

disclosure regulation in
developed countries is
likely to be determined
by more internal factors,
whereas that of
developing countries by

Level of aggregation Scaling level

Method of analysis

21

Countries

Cluster

Results, cluster denition

4/3, highly
regulated

Support hypotheses,
UK/US ``leader'' in

to under
regulated

nancial disclosure
regulation

Various
because
hierarchical

Support hypothesis,
substantial
dierences in
accounting systems
can be
explained by

more external factors


Salter and Doupnik Test the relationship
(1992) USA
between legal system
groupings (David/Brierly)
and accounting practices

Accounting
practices
(100)

50

reference to a
country's laws
(code versus case
law)
Doupnik and Salter Empirically classify current
(1993) USA
nancial reporting systems +
test the validity of a
classication proposed by

Accounting
practices
(100)

Modied PW 79 Various
+ self-generated

Cardinal

Cluster analysis
(average linkage
+ ward)

50

Various
because
hierarchical

Nobes (1983), Berry (1987)

Conrm Nobes's
classication (few
deviations), support
existence of a Latin
American group
suggested by Berry

Doupnik and Salter Test the relationship


(1995) USA
between environmental
factors and accounting
practices based on
classications

Accounting
practices
(100),
socio-economic
factors

Modied PW 79 Various
+self-generated

Cardinal

Cluster analysis
(average linkage
+ ward) (see study
from 1993)
canonical correlation

Own investigation
(1998)

Accounting
rules

Modied
Uniform
TRANSACC
Reference Matrix

Ordinal !
dichotomised

Cluster analysis
(average,
complete,
single linkage),
multidimensional
scaling

50

Various
because
hierarchical

Does not support


hypothesis, clusters
see 1993

15

Various
because
hierarchical

For 4-cluster
solution: European,
North-American,
JAP/SWE/ESP,
Australia

analysis
Find a hierarchical
classication based
on accounting rules

A. d'Arcy / Accounting, Organizations and Society 26 (2001) 327349

Classication criterion

A. d'Arcy / Accounting, Organizations and Society 26 (2001) 327349

341

Fig. 1. Dendrogram average linkage between groups.


Table 3
Dierences in the cluster solutions according to the average linkage between groups method
Method

Main eects

Dierences

Single linkage

Chaining

Dierent fusion sequence for the Netherlands,


Switzerland, Denmark, UK within the European cluster

Complete linkage

Dierent fusion sequence for Denmark within the


European cluster, the separation of Japan/Spain/
Sweden is more obvious

Average linkage within groups

Chaining

Dierent fusion sequence for the Netherlands,


Switzerland, Denmark, UK within the European cluster

the second. This group may be labelled as North


American. Consequently, an Anglo-American
cluster including the UK and the US cannot be
found. As expected from the analysis of the SM
coecients, Australia appears to have an outsider
position and is the last system to merge with the
North American cluster.
3.3. Tests of robustness
Applying the other clustering techniques as
described above, no material changes in the cluster
solutions appear. Basically, the 4-cluster-solution
is identical. Table 3 describes the main dierences.
Another test of robustness is the variation of the
similarity coecient. Basically, the dierent coef-

cients can be characterised by dierent weights


for matches and non-matches. The Rogers and
Tanimoto coecient (RT) weighs the non-matches
two times. Vice versa, the Sokal and Sneath 1Coecent (SS1) gives a double weight to the matches
so that the values will increase. The fusion process
is identical to the SM coecient and the SS1. For
the RT the Netherlands and UK are the rst to
merge with the European cluster.
This marked robustness solves another problem
that belongs to the question of dierent weightings
for the accounting methods under review. Taking
the general formula for similarity measures into
account, the RT and SS1 could be interpreted in a
dierent way: They stand for simulations of the
SM coecient, the extreme condition under which

342

A. d'Arcy / Accounting, Organizations and Society 26 (2001) 327349

all matches (for SS1) or all non-matches (for RT)


are doubled. If matches and mismatches are partly
doubled, compensatory eects can be observed.
It is possible to demonstrate that the SS1 is the
upper limit and that the RT is the lower limit of all
possible constellations of doubling certain variables applying the SM coecient.8 Because the
RT and the SS1 lead to comparable cluster results,
all kinds of weightings not exceeding the weighting
factor 2 do not aect the actual ndings. Besides
taking this robustness into account an extreme
reaction according to weighting factors >2 seems
unlikely.
Finally, the cophenetic correlation for the solution applying the Average Linkage between
Groups method in combination with the SM
coecient amounts to 0.8575. This shows a sucient degree of conformity between the fusion
measures and the original similarity coecients
(Aldenderfer & Blasheld, 1984, p. 62). The Complete Linkage method leads to a correlation of
0.6715 only. This very low value can be explained
by the late fusion of the three ``separators'' Japan/
Spain/Sweden. When thinking of this late fusion
we have to bear in mind that some similarity
measures of the three systems in respect to other
European systems are relatively high. As a result,
the qualication of Japan/Spain/Sweden as the
more heterogeneous part of the European cluster

seems more plausible than a clear separation from


the European group or the association with the
North-American cluster. In conclusion, the ndings can be summarised as follows:
(1) using the TRANSACC-sample, a European,
a North-American (including the rules of the
IASC) and an Australian cluster can be
identied for the 3-cluster-solution;
(2) Australia has a clear outsider position;
(3) the British system belongs to the European
cluster and separates clearly from the NorthAmerican group;
(4) Germany and Austria are the most homogeneous subgroup;
(5) the French and the US system are very dissimilar;
(6) compared with the North-American cluster,
the European cluster is relatively homogeneous;
(7) in-between the European cluster, a continental European core can be identied
which consists of Austria, Germany, France
and Belgium; and
(8) Spain, Sweden and Japan show the tendency
to separate from the European cluster for the
high cluster solutions (4 and more groups).
Consequently, they are the more heterogeneous part of the European cluster.

4. Multidimensional scaling

2m
xm
5
m 2m y  A m x  m
m
RT
5
2A m m
2m
xm
m
$ SS1
5
5
RT
A m y  A x ym 2A m
SS1

and 25x, y51 and A5m50 with m=sum of matches;


A=sum of all variables (hence sum of mismatches=A m); and
x, y=weighting factors 2 [1; 2].
Replacing xm through z, the rst deduction shows positive
results under all conditions, so, the function itself is increasing:
fz yA zmz
) f 0 z

yA

> 0!

1
z 1
yA m z

m z yA m z2
. . .2

A major disadvantage of all types of cluster


techniques is that the structure is forced out from
the specic classicatory technique irrespective of
whether the data shows a tendency to form clusters at all. To test the adequacy of the cluster
solutions already found it seems adequate to apply
such methods that do not lead systematically to
specic structures or to clusters at all (Gordon,
1993/1981, p. 80). Graphical methods can be
applied to picture group structures in interpreting
distances or similarities in geographical distances
of various dimensions. The nonmetrical multidimensional scaling (MDS) will be used in the
following. This method makes use of the ordinal

A. d'Arcy / Accounting, Organizations and Society 26 (2001) 327349

information of the observed similarities or distances only so that the spatial structure can be
displayed relatively independent of the chosen
distance or similarity measure (Bacher, 1996, pp.
7476).
The SM coecients are used in the following as
an input of the MDS only because this classical
variant of the method is invariant against monotone transformations. Fig. 2 shows the result of
the 2-dimensional solution. The STRESS value
a quality measure for the adequacy of the conguration after Kruskal (1964, p. 3) amounts to
0.16672. So, the picture suciently depicts the
dissimilarities of the national accounting systems
under review. We note that the same groups as in
the cluster analysis can be identied. But the micro
structure within the groups does not correspond
with the SM solutions, e.g. Switzerland and UK
are very close. This contradiction seems to be
a particular outcome of the ordinal use of the
distance measures. Typically, the MDS clearly
depicts the macrostructure but not the microstructures (Gordon, 1993/1981, p. 137).
Although the results correspond with the cluster solutions one important question remains

343

unanswered: how can the dierences in the


national accounting systems and the IAS be
explained? This work is far from giving an allembracing answer. But according to the MDS-results
some hints can be derived from the interpretation of
the two dimensions that are represented by the
axis of the two-dimensional solution. A plausible
interpretation of the axis is based on typical characteristics that cause the main dierences in the
clusters of national accounting systems.
In the following each average of all the variables
of the four groups Europe, North America,
Separator and Australia is interpreted as typical
proxy of each cluster. In the strict sense of the
word, an average of a dichotomised data set does
not exist. But values that reach nearly 1 or 0 indicate that certain accounting methods are used
homogeneously within a group. These methods
are important for the separation of two groups
that show opposing attributes. Also, they determine the kind of dimensions, which result because
the opposing attributes cause the distances pictured in the MDS solution. For the analysis, the
dierences in the groups' averages are established
according to each variable. The higher the value

Fig. 2. Multidimensional scaling 2 dimensional solution.

344

A. d'Arcy / Accounting, Organizations and Society 26 (2001) 327349

the more dierence there is and the more relevant


is the variable for the interpretation of the dimensions. In the following a dierence is assumed to
be typical if the dierence of two averages reaches
or exceeds 75%.
For the interpretation of the horizontal dimension it is mainly the dierences between the European and the North-American clusters, which are
substantial as they obviously cause a spatial differentiation in horizontal distances. When analysing the accounting methods that form the main
dierences it appears that some topics are more
signicant than others. In 13 out of 20 cases
accounting methods according to the valuation
of related companies are involved. Particularly,
variables concerning goodwill accounting, the
application of the equity-method or the full consolidation set are aected. Obviously, the NorthAmerican accounting systems lay more stress
upon the market-oriented valuation of subsidiary
companies and associated companies than the
European systems. Also, exemptions according to
the denition of a subsidiary or the choices in the
goodwill valuation are only usual in the European
group. Summarising the results, the rst dimension could be labelled the ``quality of the presentation of the participation structure''.
Taking the positions within the European cluster into account also seems plausible. The British,
the Swedish, the Danish and the Spanish systems
are positioned clearly to the right of the continental core. This corresponds with the traditional (for UK) or newly invented (for Sweden,
Denmark, Spain) group orientation of the
national accounting systems (in contrast to the
concentration on individual accounts).
At rst glance, the position of the Swedish system
is surprising. Traditionally, Sweden was characterised as a country with a strong link between
tax accounting and nancial reporting (Rundfeld,
1995, p. 2391). This inuence is still present but
some drastic changes have occurred. Most importantly, a new standard setting body, the Swedish
Financial Accounting Standards Council (Redovisningsradet) was founded in 1989. The rst
recommendation, published in 1991, on the subject
of group accounting has introduced important
changes in consolidation accounting. It is essentially

based on the related IAS (Ostman & Schuster,


1995, p. 1448).
Correspondingly, Spain used the implementation of the EU accounting directives in 1989 to
make far-reaching changes. Thus the Spanish
reduced the traditional link between taxation and
nancial accounting (Lopez Daz & Rivero
Torre, 1995, pp. 22102212). This separation
opened up the possibility of introducing a more
equity-oriented view of accounting, focusing on
shareholder and creditor protection. Also, the
national securities and exchange commission
(Comision Nacional del Mercado de Valores)
issues circulars with binding accounting standards
for quoted companies (Lopez Daz & Rivero
Torre, 1995, p. 2198). These inuences make the
position of the Spanish and Swedish system
understandable.
Accordingly, the horizontal orientation of Australia as a system of high quality at a level comparable to the North-American cluster seems
plausible taking the historical developments in
Australian accounting into account.
The interpretation of the horizontal axis as an
adequate presentation of the participation structure allows the association with a well-known
external criterion that could explain the dierences
in accounting systems. Interpreting a marketorientated and adequate presentation as an indicator for an orientation towards capital market
users the rst dimensions could stand for the level
of the capital market orientation of a national
accounting system.
The interpretation of the second dimension is
more complex. To eliminate the eects of the horizontal dimension only those dierences that
appear in the dierentiation between Europe and
Australia as well as between North America and
Australia are analysed. The dierences consist
mainly of denitions of certain types of provisions
and the rules according to the full consolidation
set. It is not possible to nd an all-embracing label
for these methods. Yet, it is apparent that Australia forbids or requires accounting methods in 9
out of 11 cases. Accordingly, the Separator group
shows a requirement or a prohibition for 7 out of
10 variables. Both groups are in the upper half of
the second dimension and hence represent the

A. d'Arcy / Accounting, Organizations and Society 26 (2001) 327349

higher level of the vertical dimension. Thus, in


summary where the vertical axis of both groups is
concerned, there is a high level of determination
interpreted as a positive relation of requests or
prohibitions compared to choices in respect to
variables that are the main cause of the dierentiation.9 Consequently, the Australian system
enforces its outsider position by certain requirements and prohibitions in contrast to accountingmethod choices. This positive relation could be
dened as a higher level of determination of groupseparating accounting methods for accounting
systems located in the upper half of the vertical
dimension. Finally, this characteristic of an
accounting system could be interpreted as the
weak or strong tendency to separate against other
national systems.
But still it has to be taken into account that
although these interpretations are convincing they
are only based on some of the variables.
5. Conclusion
The core questions of this work cannot be
answered with a simple ``yes'' or ``no''. The subject
of interest the national accounting system
itself seems too complex. The conceptional and
methodological pluralism found in accounting
classication attempts also indicates the high level
of complexity of an accounting system. The
authors reduce this complexity in dierent ways so
that diverging results do not astonish. One has to
consider that every construction of groupings
reects the author's own prejudice about the discipline (Roberts, 1995, p. 641). As a consequence,
the new classication attempt provided here can
be understood as a complementary approach
which may provide additional information of
some of the characteristics of national accounting
systems. In this relative sense some answers are
given to the following questions:

9
Relation for Australia/North-America: 17:6; Australia/
Europe: 33:15.

345

1. Is it suitable to subsume an Anglo-American


versus a continental European accounting
model just by looking purely at nancial
reporting requirements?
2. How can the role of the IAS and US-GAAP
be interpreted in relation to the national
accounting systems?
3. What impacts have the ndings on the harmon-isation process of national accounting
systems?
The most homogeneous results can be found for
the European accounting systems. Essentially,
Germany, Austria, France and Belgium can be
interpreted as the core of the continental European cluster. So far, the European accounting
directives have led to more homogeneity at the
regulation level. But there are some European
systems like Sweden or Spain that tend to make a
break from European accounting traditions. But
even if they adopt accounting methods that better
serve the information needs of capital market
participants they do not merge with the NorthAmerican cluster. Rather, they nd new combinations of accounting methods.
Against this background we are immediately
aware that it is not at all possible to nd an AngloAmerican cluster. However a relative heterogeneous North-American cluster including the
IASC can be identied. The picture of an AngloAmerican accounting model that is subsumed
specially by opponents of the harmonisation process cannot be established. But many systems that
normally come under the label ``Anglo-American''
show the tendency to concentrate on more capitalmarked-orientated accounting methods than the
European core countries. The conjecture arises
that there are no simple typical accounting methods
that stand for the capital marked orientation of an
accounting system. In fact, another analysis
applying the TRANSACC reference matrix as a
database shows that according to the degree of
determination of accounting methods an AngloAmerican cluster can be identied (d'Arcy, 2000).
Consequently, not only specic accounting methods but also the fact that methods are requested or
forbidden versus the possibility of accountingmethod choice seems to be relevant.

346

A. d'Arcy / Accounting, Organizations and Society 26 (2001) 327349

However, the analysis demonstrates that the


adoption of IAS or US-GAAP as alternative
accounting rules for European countries implies a
substantial change for the accounting users. It is
notable that particularly the European core countries allow the alternative application of IAS or
US-GAAP whereas ``Anglo-American'' systems
like the Australian or British do not accept alternative accounting principles (Knorr, 1998).
Moreover, the EU (Commission of the European Communities, 2000) has announced that it
requires IAS for all listed companies by 2005. But
the ``blind'' adoption of another accounting philosophy does not seem to guarantee the successful
reformation of a national system. At least, it is
questionable that the simple acceptance of other
standards will improve the quality of the national
accounting practices automatically. Spain and
Sweden have already started to solve this problem
by developing an individual solution for a capital
market-orientated accounting system. Germany has
already reacted by establishing the German
Accounting Standards Committee (Deutscher Standardisierungsrat) in May 1998. Thus, it appears
that accounting is a more or less a local/national
discipline (Lukka & Kasanen, 1996, p. 757).
However this discipline tends to be orientated on
international discussions and developments, too.
Analysing this process it becomes obvious that a
capital market-orientated accounting system does
not develop automatically. The ndings indicate
that especially in a changing environment, the
foundation of new standard-setting bodies seems
to form important landmarks in improving the
national accounting system. Besides, the reactions
of the EU, the IASC and the SEC in respect to the
local activities are important.
Acknowledgements
The author gratefully acknowledges helpful
comments by Prof. Dieter Ordelheide, Prof. Ulrich
Rendtel, Prof. Simon Archer, Prof. Christopher
Nobes, Nina Poppinga, and two unknown
reviewers. Special thanks to Katie Meyers and
Cynthia Awuku for checking my English-language
usage.

References
Aldenderfer, M. S., & Blasheld, R. K. (1984). Cluster analysis,
(Sage University Paper series on Quantitative Applications in
the Social Sciences, Series No. 07-001). Beverly Hills/London: Sage Publication.
Alexander, D., & Nobes, C. W. (1994). A European introduction
to nancial reporting. London: Prentice-Hall International.
Alexander, D., & Archer, S. (2000). On the myth of ``AngloSaxon'' nancial accounting. International Journal of
Accounting (in preparation).
AlNajjar, F. K. (1986). Standardization in accounting practices,
a comparative international study. International Journal of
Accounting, 21, 161176.
American Accounting Association (AAA) (1977). Report of the
American Accounting Association Committee on International Accounting Operations and Education. Accounting
Review (AAA), 52(Suppl), 65132.
Amir, E., Harris, T. S., & Venuti, E. K. (1993). A comparison
of the value-relevance of U.S. versus non-U.S. GAAP
accounting measures using form 20-F reconciliations. Journal of Accounting Research, 31(Suppl), 230264.
Archer, S., & McLeay, S. (1992). European nancial reporting:
issues of denition and classication of national systems
of nancial accounting. Paper presented at the Annual
Congress of the European Accounting Association. Madrid,
Spain.
Archer, S., Delvaille, P., & McLeay, S. (1995). The measurement of harmonisation and the comparability of nancial
statement items: within-country and between-country eects.
Accounting and Business Research, 25, 6780.
Bacher, J. (1996). Clusteranalyse (2nd ed.). Munchen/Wien:
Oldenbourg.
Barrett, M. E. (1977). The extent of disclosure in annual
reports of large companies in seven countries. International
Journal of Accounting, 12, 125.
Baydoun, N., & Willett, R. (1995). Cultural relevance of western accounting systems to developing countries. ABACUS,
31, 6792.
Belkaoui, A. R. (1989). Cultural determinism and professional
self-regulation in accounting: a comparative ranking.
Research in Accounting Regulation, 3, 93101.
Belkaoui, A. R. (1994). International and multinational
accounting. London: The Dryden Press.
Berry, I. (1987). The need to classify worldwide practices.
Accountancy, 100, 9091.
Bloom, R., & Naciri, M. A. (1989). Accounting standard setting and culture: a comparative analysis of the United States,
Canada, England, West Germany, Australia, New Zealand,
Sweden, Japan, and Switzerland. International Journal of
Accounting, 24, 7097.
Bloom, R., Long, S., & Collins, M. (1994). Japanese and
American accounting: explaining the dierences. In K. Most,
Advances in international accounting (pp. 265284). London:
JAI Press.
Bockem, H., & d'Arcy, A. (1999). Evolution of (international)
accounting systems critical assessment of the environ-

A. d'Arcy / Accounting, Organizations and Society 26 (2001) 327349


mental determinism theory with an application to tax inuence. Zeitschrift fur betriebswirtschaftliche Forschung (ZfbF),
51, 6076.
Boczo, A. (1996). Culture, socialisation, and mutual recognition, critical perspectives on the mutual recognition of professional accountancy qualications in the European Union,
Paper presented at the Annual Congress of the European
Accounting Association. Bergen, Norway.
Choi, F. D. S., & Mueller, G. G. (1992). International accounting (2nd ed). Englewood Clis: Prentice-Hall International.
Commission of the European Communities (2000). Communication from the Commission to the Council and the European
Parliament: EU Financial Strategy: the way forward, COM
(2000) 359 nal.
Cooke, T. E. (1992). The impact of size, stock market listing
and industry type on disclosure in the annual reports of
Japanese listed corporations. Accounting and Business
Research, 22, 229237.
Cooke, T. E., & Wallace, R. S. O. (1990). Financial disclosure
regulation and its environment: a review and further analysis. Journal of Accounting and Public Policy, 9, 79110.
d'Arcy, A. (1999). Gibt es eine anglo-amerikanische oder eine
kontinentaleuropaische Rechnungslegung? Klassen nationaler Rechnungslegungssysteme zwischen Politik und empirischer Uberprufbarkeit. Dissertation. Frankfurt: Peter Lang.
d'Arcy, A. (2000). The degree of determination of national
accounting systems an empirical investigation. Schmalenbach Business Review, 53, 4567.
Da Costa, R. C., Bourgeois, J. C., & Lawson, W. M. (1978). A
classication of international nancial accounting practices.
International Journal of Accounting, 13, 7385.
Daley, L. A., & Mueller, G. G. (1989). Accounting in the arena
of world politics, crosscurrents of international standardsetting activities. In J. N. Sheth, & A. Eshghi, Global
accounting perspectives (pp. 2132). Cincinnati: South-Western Publishing Co.
Day, R. (1996). Public interest: private power, Paper presented
at the Annual Congress of the European Accounting Association. Bergen, Norway.
Doupnik, T. S. (1987). Evidence of international harmonization of nancial reporting. The International Journal of
Accounting Education and Research, 23, 4767.
Doupnik, T. S., & Salter, S. B. (1993). An empirical test of a
judgemental international classication of nancial reporting
practices. Journal of International Business Studies, 24, 4160.
Doupnik, T. S., & Salter, S. B. (1995). External environment,
culture, and accounting practice: a preliminary test of a general
model of international accounting development. International
Journal of Accounting, 30, 189207.
Emenyonu, E. N., & Gray, S. J. (1992). EC-harmonisation:
an empirical study of measurement practices in France,
Germany and the UK. Accounting and Business Research, 23,
4958.
Epps, R. W., & Oh, J.-W. (1997). Market perception of foreign
nancial reports: dierential earnings response coecients
between U.S. and foreign GAAP. Journal of International
Accounting, Auditing & Taxation, 6, 4974.

347

Fechner, H. E., & Kilgore, A. (1994). The inuence of cultural


factors on accounting practice. International Journal of
Accounting, 29, 265277.
Flower, J. (1997a). The future shape of harmonization: the EU
versus the IASC versus the SEC. European Accounting
Review, 6, 281303.
Flower, J. (1997b). Introduction. In J. Flower, & C. Lefebvre,
Comparative studies in accounting regulation in Europe (pp.
2144). Leuven: Acco.
Frank, W. G. (1979). An empirical analysis of international
accounting principles. Journal of Accounting Research, 17,
593605.
Goebel, A. (1995). Moglichkeiten und Probleme einer Anwendung der International Accounting Standards, Anpassungsstrategien fur Deutsche Konzerne. Deutsches Steuerrecht, 33,
10371039.
Goodrich, P. S. (1982). A typology of international accounting
principles and policies. AUTA Review, 14, 3761.
Gordon, A. D. (1993/1981). Classication, methods for the
exploratory analysis of multivariate data. London: Chapman
& Hall.
Gray, S. J. (1988). Towards a theory of cultural inuence
on the development of accounting systems internationally.
Abacus, 24, 115.
Gray, S. J., Campbell, L. G., & Shaw, J. C. (1984). International nancial reporting a comparative international survey
of accounting requirements and practices in 30 countries.
Basingstoke: Macmillan.
Hagigi, M., & Sponza, A. (1990). Financial statement analysis of Italian companies: accounting practices, environmental factors, and international corporate performance
comparisons. International Journal of Accounting, 25, 234
251.
Haskins, M. E., Ferris, K. R., & Selling, T. I. (1996). International nancial reporting and analysis, a contextual emphasis.
Chicago: Irvin.
Hateld, H. R. (1966/1911). Some variations in accounting
practice in England, France, Germany and the United States.
Journal of Accounting Research, 4, 169182. (Reprint from
1911).
Herrmann, D., & Thomas, W. (1995). Harmonisation of
accounting measurement practices in the European Community. Accounting and Business Research, 25, 253265.
Hoarau, C. (1995). International accounting harmonization,
American hegemony or mutual recognition with benchmarks? European Accounting Review, 4, 217233.
Hofstede, G. (1980). Culture's consequences: international differences in work-related values. Beverly Hills: Sage Publications.
Hofstede, G., & Schreuder, H. (1987). Reply, a joint reply to
Montagna. In B. E. Cushing, Accounting and Culture, Plenary Session Papers and Discussants' Comments from the 1986
Annual Meeting of the American Accounting Association (pp.
2930). Beverly Hills: Sage Publications.
Joos, P., & Lang, M. (1994). The eects of accounting diversity: evidence from the European Union. Journal of
Accounting Research, 32(Suppl), 141168.

348

A. d'Arcy / Accounting, Organizations and Society 26 (2001) 327349

Knorr, L. (1998). Europe opens to IAS. IASC Insight, March, 1.


Kruskal, J. B. (1964). Multidimensional scaling by optimizing
goodness-of-t to a nonmetric hypothesis. Psychometrika,
29, 127.
Lopez Daz, A., & Rivero Torre, P. (1995). Spain, individual
accounts. In D. Ordelheide, & KPMG, Transnational
Accounting (TRANSACC) (pp. 21842303). Basingstoke:
McMillan.
Lukas, W. (1995). Austria. In D. Alexander, & S. Archer,
Miller European accounting guide (pp. 10211065); 2nd ed.
San Diego: Harcourt Brace.
Lukka, K., & Kasanen, E. (1996). Is accounting a global or a
local discipline? Evidence from major research journals.
Accounting, Organizations and Society, 21, 755773.
MacArthur, J. B. (1996). An investigation into the inuence
of cultural factors in the lobbying of the International
Accounting Standards Committee. International Journal of
Accounting, 31, 213237.
Mathews, M., & Perera, H. (1991). Accounting theory and
development. London: Chapman Hall.
Meek, G. K., & Saudagaran, S. M. (1990). A survey of research
on nancial reporting in a transnational context. Journal of
Accounting Literature, 9, 145182.
Montagna, P. D. (1987). Discussion comments. In B. E. Cushing, Accounting and Culture, Plenary Session Papers and
Discussants' Comments from the 1986 Annual Meeting of the
American Accounting Association (pp. 2328). Beverly Hills:
Sage Publications.
Mueller, G. G. (1967). International accounting. New York:
Macmillan.
Mueller, G. G. (1968). Accounting principles generally accepted in the United States versus those generally accepted elsewhere. International Journal of Accounting, 3, 91103.
Nair, R. D. (1982). Empirical guidelines for comparing international accounting data. Journal of International Business
Studies, 13, 8598.
Nair, R. D., & Frank, W. G. (1980). The impact of disclosure
and measurement practices on international accounting
classication. Accounting Review, LV, 426450.
Needles, B. E., Powers, M., & Revsine, L. (1991). Financial
disclosures and institutional characteristics: pension reporting dierences across six countries. International Journal of
Accounting, 26, 190205.
Nobes, C. W. (1981). An empirical analysis of international
accounting principles: A Comment. Journal of Accounting
Research, 19, 268270.
Nobes, C. W. (1983). A judgemental international classication
of nancial reporting practices. Journal of Business, Finance
and Accounting, 10, 119.
Nobes, C. W. (1992). International classication of nancial
reporting (2nd ed.). London: Routledge.
Nobes, C. W., & Matatko, J. (1980). Classication of national
systems of nancial accounting. AUTA Review, 12, 5778.
Nobes, C. W., & Parker, R. (1995). Comparative international
accounting. New York: Prentice Hall International.
Norus is, M. (1994). SPSS professional statistics 6.1. Chicago:
SPSS Inc.

Ordelheide, D. (1997). Regulierung der rechnungslegung


Okonomische `Zwange' und kulturelle unterschiede. In H.
Engelhardt, Interkulturelles Management Theoretische
Fundierung und funktionsbereichsspezische Konzepte (pp.
235259). Wiesbaden: Gabler.
Ordelheide, D., & KPMG (Eds), (1995). Transnational
accounting (TRANSACC). Basingstoke: McMillan.
Ordelheide, D., & Semler, A. (1995). A reference matrix. In D.
Ordelheide, & KPMG, Transnational accounting (TRANSACC) (pp. 167). Basingstoke: McMillan.
Ostman, L., & Schuster, W. (1995). Sweden, group accounts. In
D. Ordelheide, & KPMG, Transnational accounting (TRANSACC), McMillan: Basingstoke, 24432481
Peller, P. R., & Schwitter, F. J. (1991). A summary of
accounting principle dierences around the world. In F. D. S.
Choi, Handbook of international accounting (pp. 4.14.23).
New York: John Wiley & Sons.
Perera, M. H. B. (1989). Towards a framework to analyze the
impact of culture on accounting. International Journal of
Accounting, 24, 4256.
Perera, M. H. B., & Mathews, M. R. (1990). The cultural relativity of accounting and international patterns of social
accounting. In K. Most, Advances in international accounting
(pp. 216251). London: JAI Press.
Previts, G. J. (1975). On the subject of methodology and models
for international accountancy. International Journal of
Accounting, 10, 112.
Price Waterhouse (1973). Accounting principles and reporting
practices: a survey in 38 countries. London.
Price Waterhouse (1975). Accounting principles and reporting
practices: a survey in 46 countries. London.
Price Waterhouse (1979). International survey of accounting
principles and reporting practices. London.
Puxty, A. G., Willmott, H. C., Cooper, D. J., & Lowe, T.
(1987). Modes of regulation in advanced capitalism: locating
accountancy in four countries. Accounting, Organizations and
Society, 12, 273291.
Radebough, L. H., & Gray, S. J. (1993). International accounting and multinational enterprises (3rd ed.). New York: John
Wiley & Sons.
Rahman, A., Perera, H., & Ganeshanandam, S. (1996). Measurement of formal harmonisation in accounting: an
exploratory study. Accounting and Business Research, 26,
325339.
Riahi-Belkaoui, A. (1997). The nature and determinants of disclosure adequacy an international perspective. Westport/
London: Quorum Books.
Roberts, A. (1995). The very idea of classication in international
accounting. Accounting, Organizations and Society, 20, 639664.
Rundfeld, R. (1995). Sweden individual accounts. In D.
Ordelheide, & KPMG, Transnational accounting (TRANSACC) (pp. 23712442). Basingstoke: McMillan.
Ryan, J. B. (1985). Book review of international classication
of nancial reporting. Accounting and Finance, November,
8788.
Salter, S. B., & Doupnik, T. S. (1992). The relationship between
legal systems and accounting practices: a classication exercise.

A. d'Arcy / Accounting, Organizations and Society 26 (2001) 327349


In K. Most, Advances in international accountings (pp. 322).
London: JAI Press.
Seidler, L. J. (1967). International accounting the ultimate
theory course. Accounting Review, 42, 775781.
Shoenthal, E. R. (1989). Classication of accounting systems
using competencies as a discriminating variable: a Great
BritainUnited States study. Journal of Business, Finance and
Accounting, 16, 549563.
Sneath, P. H. A., & Sokal, R. R. (1973). Numerical taxonomy,
the principles and practice of numerical classication. San
Francisco: W.H. Freddman and Company.
Taylor, P., & Turley, S. (1986). The regulation of accounting.
Oxford: Basil Blackwell.
Terberger, E. (1994). Neo-institutionalistische ansatze. Wiesbaden:
Gabler.

349

van der Tas, L. G. (1995). International accounting harmonization, American hegemony or mutual recognition with
benchmarks? A comment. The European Accounting Review,
4, 255260.
Walton, S. (1992). Harmonization of accounting in France and
Britain: some evidence. ABACUS, 28, 186199.
Watts, R. L., & Zimmerman, J. L. (1986). Positive accounting
theory. Englewood Clis: Prentice-Hall International.
Weetman, P., & Gray, S. J. (1991). A comparative international
analysis of the impact of accounting principles on prots:
the USA versus the UK, Sweden, and The Netherlands .
Accounting and Business Research, 21, 363379.
Ze, S. A. (1972). Forging accounting principles in ve countries:
a history and an analysis of trends. Campaign: Stipes Publishing.

Vous aimerez peut-être aussi